498 CONTRIBUTIONS TO OUR KNOWLEDGE OF SOIL-FERTILITY, ii., 



No. 5 of the series is apparently accidentally low, and an 

 allowance should be made for it A^ariations in the reaction of 

 the medium do not appear to have much influence upon the 

 jfrowth of Rhizobial forms, so long as the small quantity of soda 

 is added juHt before the ])Iates are poured. The addition of the 

 soda to the bulk of the medium in its preparation, has no influence 

 in assisting the growth of the bacteria. This curious time-etfect 

 is seen when Expt. iv. is compared with Expt. vi. 



The Influence of Carbonate of Soda. 



When the soda is added to the molten agar just before pouring 

 the plates, the medium becomes so selective as to be almost 

 specific. On the plates of Expt. vi., which is typical of the 

 previous experiments, the Rhizobial colonies constituted from 

 33% to 86% of the bacterial colonies. 



Experiment vi. was repeated with the following results: — 



